The 2017 TUFLOW release notes mention that input grid extents are checked when using the "Read GRID ==" command. If the extent is outside the 2D domain the grid is skipped to reduce simulation startup.
I have noticed that for quadtree models with nesting, that inputs grids are processed for each refinement level even if some grids do not fall in the refinement region. For example, a quadtree model with base 10m grid size and one 5m refinement area using DEM tiles to set Zpts. Each tile in the DEM will be processed to the 5m refinement level (GRID_TILE_X.5m.xf4) even if that tile does not fall within any part of the 5m refinement region.
Does this unnecessarily increase startup times by generating xf files that aren't needed?